loading
loading
After a decade building and scaling product-led startups across fintech and insurance, Prateek Jogani set out in 2025 to find his next problem to solve. AI was the unavoidable frontier, and a closer look at the ecosystem revealed a disconnect between industry hype and on-the-ground reality. Speaking to leadership teams across mid-cap, growth-stage, and pre-IPO enterprises, Jogani noticed a recurring pattern in discussions. Almost nobody had AI doing meaningful work inside the business. The models were undeniably powerful, yet enterprises continued to hit roadblocks. Structuring enterprise data proved messy, governing autonomous workflows was complex and costly, and teams were ill-equipped for the organisational overhaul AI required. At the same time, legacy IT services firms, the traditional stewards of enterprise technology, struggled to keep pace with AI and were absent from those conversations. Seeing an execution vacuum, Jogani set out to create a lean partner that could embed AI directly into enterprise workflows using AI-native engineers and modern tooling. Backed with $6.6 million in funding from Paytm CEO Vijay Shekhar Sharma, CRED Founder Kunal Shah, AEOS Labs Founder Varun Mayya, and a cohort of VCs, Jogani this month launched Parallel Loop—positioned not as an advisory consultancy, but as an AI enablement and execution firm. This is now a growing genre of companies in India. Under one umbrella, they can be safely called ‘boutique’ AI service firms. One misconception that has spectacularly fallen apart in the generative AI era is that AI will eliminate the need for engineers. Deploying frontier models and agents still requires teams to integrate them with proprietary data, manage risks and control costs. As a result, India, the world’s IT services capital, may not be facing the bleak future once predicted. Although the industry faces challenges such as the erosion of traditional hourly billing models, dilemmas in hiring entry-level employees whose roles AI can perform, and declining quarterly hikes, demand for services remains strong. TCS reported annualised AI revenue of $3.1 billion in the September quarter, along with $9.6 billion in total contract value. HCLTech, meanwhile, reported $3.26 billion in new deal wins, with annualised advanced-AI revenue rising 28% sequentially to $219 million. As large, established firms tap into the AI opportunity and modernise their service models for the AI era, a parallel ecosystem of boutique AI consulting companies is emerging. Even as traditional IT services firms aim to bring AI closer to customers and speed up enterprise deployments, AI-native consultancies are positioning themselves as smaller, faster engineering partners that can take AI projects from concept to production in days, not months. Enter Boutique AI Consultancy Firms Parallel Loop’s advantage, Jogani argues, lies in combining AI-native engineers with tools that let small teams accomplish more. “What works in our favour is that we are all AI native engineers and practitioners by default,” Jogani tells AIM. “We have the luxury of starting in 2026, so ground up, everything is AI native.” Rather than staffing each project with a large team to write code, the firm uses a “software factory” of AI agents, with engineers building and integrating the resulting workflows. “Recently, one of our customers received a quote from a services company for five months—we delivered the app in 10 days,” he says. “Because of our size, we can move fast.” Jogani gives the example of Jev, a new decision AI model that doesn’t output text but instead provides probabilities for which outcome would work best. This approach can be significantly more efficient than text-based models for some use cases. “Our workflows already run Jev in production for our enterprises,” says Jogani, highlighting how quickly the company can implement the latest AI frameworks and techniques. Parallel Loop deploys its AI stack within the customer’s own infrastructure rather than hosting enterprise workloads on its own cloud. This lets enterprises keep control of their data and deployment environment while using Parallel Loop’s AI stack to connect existing systems and build automated workflows. Its commercial model combines a monthly licence fee for the platform, packages of forward-deployed engineering hours, one-time software development fees, and recurring maintenance charges. Make no mistake, these companies are not on a mission to replace SaaS platforms by building new ones for their clients. Jogani reiterated that enterprise software such as SAP and Salesforce already serve as systems of record, handling critical business data and compliance requirements. Adding AI capabilities to individual platforms risks preserving the silos between them. Instead, Parallel Loop connects these systems through a horizontal AI layer, letting businesses build workflows that draw on data across applications and departments. Selling Outcomes, Not Specifications “Traditional IT services typically deliver against a specification,” said Akhil Verghese, Co-Founder and CEO of Krazimo, another boutique AI consulting firm. “We sell an outcome, using AI and AI-assisted development to achieve it.” Verghese explains that this might mean automating a workflow; building an agent that performs a task as well as, or better than, an employee and scales with fluctuating demand; or making a company’s information, tools, and APIs accessible to employees and AI systems. He also points to the advantage of using cutting-edge AI frameworks to achieve goals quickly. “It changes how much a small team can accomplish, and using these systems ourselves gives us direct experience of what clients need to make them dependable.” The company already generates $500,000 per employee. For exploratory projects, Krazimo offers a cancellable monthly subscription. For projects with clearly defined requirements and measurable goals, it charges a fixed price tied to delivery milestones. Clients can also exit without payment during the first 30 days if they are unhappy with the work. Rather than billing purely for engineering hours, Krazimo structures engagements around deliverables and measurable business goals. It says clients typically break even within three to six months, with profits increasing by an average of 17%. Similarly, Parallel Loop, with an outcome-based delivery model, has revealed securing eight enterprise AI clients within months of incorporation, with $2 million in booked revenue. But can large IT firms in India move with this model at scale? “Boutique AI firms are strongest in areas where incumbents would have to cut their own revenue to compete,” Leslie Joseph, VP and Principal Analyst, Forrester, tells AIM. India’s large IT services firms have traditionally relied on a mix of time-and-materials contracts, which bill for staff and hours deployed, and fixed-price contracts for an agreed scope of work. Neither model necessarily ties payment to business outcomes. For example, Infosys reported that fixed-price contracts accounted for 58% of its revenue in FY2026. Wipro, meanwhile, derived 45.6% of its revenue from fixed-price and volume-based contracts in FY2026, with the remaining 54.4% coming from time-and-materials contracts. However, outcome-based pricing is gaining ground as AI changes delivery economics, but remains a minority model. In August, Coforge, a Tier 2 Indian IT company, revealed that outcome-based contracts contributed 6-7% of its $2.5 billion revenue run rate. TCS has also reported a shift towards outcome-based commitments, particularly in agentic global business services “AI shrinks the billable work in these contracts, and clients now expect the savings passed back to them,” says Joseph. Boutique AI firms, by contrast, do not carry the same legacy contractual commitments. Instead, they make a compelling promise: to automate much of the routine work incumbents bill for. Joseph also explains that these firms can staff engagements with more senior engineers, positioning themselves as more AI-native than their larger competitors. Besides, outcome-based pricing is also viable for these firms in narrow workflows where results are easy to measure. “That model remains unproven for large transformation programs,” says Joseph. How to Create a Boutique AI Consulting Firm Many boutique AI companies are founded by engineers with direct experience developing and scaling software. They also use their personal experience and skills as a key selling point to attract clients. Verghese spent six years at Google, where he worked on Gemini projects within Workspace and helped develop techniques to reduce hallucinations and improve AI reliability. “Both (Verghese and his Co-Founder, Mridul Nagpal) of us loved working at Google, so we didn’t take this step lightly,” says Verghese. “Google set a particularly high bar for the reliability of these agents. In attempting to meet that bar, we adopted many techniques that we weren’t seeing people talk about publicly.” With Google's permission, Verghese and Nagpal began helping companies seek their advice and expertise. “We continued running the business while maintaining our roles at Google for over a year and a half. We promised ourselves that when the company hit $50k in MRR, we’d leave,” he says. “That happened in June of 2025, and by July, both of us had left.” Similarly, Jogani brings nearly two decades of experience and served as CTO of several tech companies across Asia before starting Parallel Loop. “My last company was listed on the Indonesian stock exchange, and we built the entire stack of it in four months,” says Jogani. However, one particularly interesting case is Cars24. The automotive logistics giant spent the last two years building and running AI agents for customer support, finance, vehicle inspections and other workflows. Rather than using AI for isolated tasks, Cars24 rebuilt parts of its business around AI-driven workflows, embedded in customer engagement, vehicle inspections, lending, engineering and internal operations. It said its AI voice agents handled 2.58 million conversation minutes in Q1 FY27, while chatbots processed more than 5,50,000 customer requests. AI-led systems conducted 80% of vehicle inspections in the quarter, up from 25% in the previous quarter. Its inspection platform reduced average inspection time from 45 minutes to 26 minutes. Similarly, AI-led loan disbursals rose to 65%, from 30% Having achieved these results internally, the company began exploring whether it could help more companies replicate its success. Thus emerged Deployment Inc. The Indian automotive sales giant has spun off its own AI consultancy firm after successfully deploying AI within its operations and now wants to replicate that success across the broader ecosystem. The company launched the firm in August, backed by an initial $5 million investment from Cars24 and co-founded by Jayesh Gupta, who led AI at Cars24, and Aayush Gupta, a former xAI engineer. “Cars24 is our first operating proof, and the experience there is what we are taking to other enterprises,” Jayesh tells AIM. “Implementing AI inside Cars24 taught us what happens after a successful demonstration: how users behave, where systems fail, how exceptions are handled and whether the economics hold.” For Deployment Inc, the initial focus is on founder-led, tech-enabled businesses, with meaningful operating scale. “We look for an important business problem, an executive sponsor, access to relevant systems and data, and an operating team that can work with us,” says Jayesh. He explains that a small team of forward-deployed engineers work closely with the customer’s operators and technology team, staying involved through discovery, building, integration, rollout and adoption. “We establish a baseline, put the workflow into production with appropriate controls, and measure whether it improves the business,” he adds. These firms can also work closely with frontier model makers in the US. Deployment Inc is an OpenAI Select Partner, a programme from the ChatGPT maker which offers access to technical resources, training and support for enterprise deployments. Deploying AI for Non-Digital-Native Firms Parallel Loop aims to serve a wider market, including companies in healthcare, retail, automotive, and manufacturing sectors that see AI as an attractive option but have limited experience in implementing it. Jogani says most leaders of such enterprises in India, which aren’t tech-forward or digital native to begin with, want to implement AI but have less capability. “This target segment helps because they have all the resources and distribution, but lack the capability of going all out,” he says. “Because every time an idea comes out—they think it takes six months to implement. I want to change that.” Havells India was one of Parallel Loop’s clients at launch. Verghese also shares a similar experience: his biggest successes have come from businesses with limited digital presence and substantial manual work. “Particularly in home services, construction and healthcare—we have identified opportunities for AI to reduce workload by up to 20% and boost revenue by over 20%,” he says. Krazimo is exploring government business in India through its partnership with OriaNode Technologies, an early-stage AI company that builds custom AI systems for Indian government departments, public sector undertakings (PSUs) and statutory authorities. AI consulting firms nationwide see this as a promising opportunity. For instance, TunerLabs, an AI engineering consultancy, helped a Karnataka-based building materials manufacturer automate accounting, payment collections, demand forecasting, and sales tracking, reporting monthly savings of lakhs of rupees. Similarly, Rooted AI, an India-based enterprise AI consultancy, says it helped a Hosur-based auto-parts manufacturer deploy computer vision across three production lines, reducing scrap by 34% in five weeks. In a separate engagement, it built an AI-powered dispatch-routing system for a Chennai-based third-party logistics provider handling more than 200 deliveries a day. By factoring in live traffic, vehicle loads and delivery windows, the system reduced fuel costs by 28%, according to the company. However, a unique set of challenges arises, particularly with potential Indian customers before the pen is put to paper. Show Me the Money “The best part about India—which at times is not great for businesses—is that they want to see ROI,” says Jogani. Some businesses in India remain unwilling to invest in AI simply because it is gaining traction globally. Instead, they want vendors to demonstrate a clear return on investment before committing to wider deployments. Even when interest exists, some are reluctant to spend, as the relatively low cost of human labour can make automation less compelling. “Price sensitivity creates pressure when other providers offer apparently similar solutions for a third of the price,” says Verghese. He explains that this difference becomes clear when the system must operate inside the business, using real customer data, exceptions and consequences. “That is where you discover whether you bought a dependable system with the right governance or an impressive demonstration from someone who has learned to prompt ChatGPT,” he says. While firms like Parallel Loop, Krazimo, and Deployment Inc. use lean teams, outcome-based pricing, and an automation-first approach that speeds deployment timelines, large IT services incumbents retain advantages of their own. Why GCCs are the Dark Horses Joseph argued that large IT services firms’ advantage rests firmly on scale, established client relationships, and control over the systems AI agents must operate through. They already manage core banking platforms, ERP systems and mainframes, giving them access to the infrastructure and integration expertise needed to deploy AI across complex enterprises. Established vendors also have an advantage in regulated sectors, where auditability, governance and accountability matter as much as model performance. For example, in financial services, the RBI’s FREE-AI framework recommends board-approved AI policies, risk classification and audit mechanisms for AI systems. These expectations add governance and documentation requirements to enterprise deployments, favouring vendors with experience in security, integration and regulated systems. “Incumbents also have the money to take over client operations and buy capability,” says Joseph. He points to TCS’s recent deals as examples. Recently, TCS agreed to take over Best Buy’s India global capability centre, bringing its employees and technology operations into TCS under a multi-year agreement to build an AI-centred innovation hub. In August, it agreed to acquire Porsche’s IT consulting subsidiary MHP for €320 million, alongside a five-year strategic partnership with Porsche worth €1.25 billion. These deals give TCS access to established teams, specialised expertise and long-term client engagements, advantages that smaller AI consultancies would struggle to replicate. Boutique AI firms will also need to grapple with talent supply and demand dynamics. Hiring competent, highly skilled AI engineers is one challenge, but retaining them, and preventing larger product firms and internal engineering teams from poaching them, is another. Moreover, frontier model companies are themselves vying for partnerships with Indian enterprises. Even without a sizable team dedicated to AI transformation, enterprises may consider working directly with OpenAI and Anthropic, and their forward-deployed engineering teams, as a more agile alternative. Moreover, Joseph said both large IT firms and smaller AI consultancy firms will have to contend with the challenges of growing internal engineering teams. Global capability centres in India hired 2,00,000 net employees in fiscal 2026, almost twice the 1,10,000 added by IT services firms. “Those centres also take work away from both groups,” says Joseph. Over the last few years, the share of companies establishing—and expanding their in-house engineering teams has steadily increased. This eliminates any opportunity for external consulting teams. Founders like Jogani also recognise the challenge of competing with large IT firms today, which are building forward-deployed engineering teams to work closely with customers, deploying frontier AI coding tools across their workforce, and modernising their service models. “I won’t assume that large IT firms will not evolve, or not be capable of doing so,” he says. “But if I have lakhs of people in my organisation, to change the way they operate will not happen in a few months, so it'll take some time, and there is a window.” But whether this window is hinging outwards or otherwise will only become clear as firms offer more clues, and as clarity emerges on AI revenues, the number and scale of projects, and how much of that business translates into measurable returns.
After a decade building and scaling product-led startups across fintech and insurance, Prateek Jogani set out in 2025 to find his next problem to solve. AI was the unavoidable frontier, and a closer look at the ecosystem revealed a disconnect between industry hype and on-the-ground reality. Speaking to leadership teams across mid-cap, growth-stage, and pre-IPO enterprises, Jogani noticed a recurring pattern in discussions. Almost nobody had AI doing meaningful work inside the business. The models were undeniably powerful, yet enterprises continued to hit roadblocks. Structuring enterprise data proved messy, governing autonomous workflows was complex and costly, and teams were ill-equipped for the organisational overhaul AI required. At the same time, legacy IT services firms, the traditional stewards of enterprise technology, struggled to keep pace with AI and were absent from those conversations. Seeing an execution vacuum, Jogani set out to create a lean partner that could embed AI directly into enterprise workflows using AI-native engineers and modern tooling. Backed with $6.6 million in funding from Paytm CEO Vijay Shekhar Sharma, CRED Founder Kunal Shah, AEOS Labs Founder Varun Mayya, and a cohort of VCs, Jogani this month launched Parallel Loop—positioned not as an advisory consultancy, but as an AI enablement and execution firm. This is now a growing genre of companies in India. Under one umbrella, they can be safely called ‘boutique’ AI service firms. One misconception that has spectacularly fallen apart in the generative AI era is that AI will eliminate the need for engineers. Deploying frontier models and agents still requires teams to integrate them with proprietary data, manage risks and control costs. As a result, India, the world’s IT services capital, may not be facing the bleak future once predicted. Although the industry faces challenges such as the erosion of traditional hourly billing models, dilemmas in hiring entry-level employees whose roles AI can perform, and declining quarterly hikes, demand for services remains strong. TCS reported annualised AI revenue of $3.1 billion in the September quarter, along with $9.6 billion in total contract value. HCLTech, meanwhile, reported $3.26 billion in new deal wins, with annualised advanced-AI revenue rising 28% sequentially to $219 million. As large, established firms tap into the AI opportunity and modernise their service models for the AI era, a parallel ecosystem of boutique AI consulting companies is emerging. Even as traditional IT services firms aim to bring AI closer to customers and speed up enterprise deployments, AI-native consultancies are positioning themselves as smaller, faster engineering partners that can take AI projects from concept to production in days, not months. Enter Boutique AI Consultancy Firms Parallel Loop’s advantage, Jogani argues, lies in combining AI-native engineers with tools that let small teams accomplish more. “What works in our favour is that we are all AI native engineers and practitioners by default,” Jogani tells AIM. “We have the luxury of starting in 2026, so ground up, everything is AI native.” Rather than staffing each project with a large team to write code, the firm uses a “software factory” of AI agents, with engineers building and integrating the resulting workflows. “Recently, one of our customers received a quote from a services company for five months—we delivered the app in 10 days,” he says. “Because of our size, we can move fast.” Jogani gives the example of Jev, a new decision AI model that doesn’t output text but instead provides probabilities for which outcome would work best. This approach can be significantly more efficient than text-based models for some use cases. “Our workflows already run Jev in production for our enterprises,” says Jogani, highlighting how quickly the company can implement the latest AI frameworks and techniques. Parallel Loop deploys its AI stack within the customer’s own infrastructure rather than hosting enterprise workloads on its own cloud. This lets enterprises keep control of their data and deployment environment while using Parallel Loop’s AI stack to connect existing systems and build automated workflows. Its commercial model combines a monthly licence fee for the platform, packages of forward-deployed engineering hours, one-time software development fees, and recurring maintenance charges. Make no mistake, these companies are not on a mission to replace SaaS platforms by building new ones for their clients. Jogani reiterated that enterprise software such as SAP and Salesforce already serve as systems of record, handling critical business data and compliance requirements. Adding AI capabilities to individual platforms risks preserving the silos between them. Instead, Parallel Loop connects these systems through a horizontal AI layer, letting businesses build workflows that draw on data across applications and departments. Selling Outcomes, Not Specifications “Traditional IT services typically deliver against a specification,” said Akhil Verghese, Co-Founder and CEO of Krazimo, another boutique AI consulting firm. “We sell an outcome, using AI and AI-assisted development to achieve it.” Verghese explains that this might mean automating a workflow; building an agent that performs a task as well as, or better than, an employee and scales with fluctuating demand; or making a company’s information, tools, and APIs accessible to employees and AI systems. He also points to the advantage of using cutting-edge AI frameworks to achieve goals quickly. “It changes how much a small team can accomplish, and using these systems ourselves gives us direct experience of what clients need to make them dependable.” The company already generates $500,000 per employee. For exploratory projects, Krazimo offers a cancellable monthly subscription. For projects with clearly defined requirements and measurable goals, it charges a fixed price tied to delivery milestones. Clients can also exit without payment during the first 30 days if they are unhappy with the work. Rather than billing purely for engineering hours, Krazimo structures engagements around deliverables and measurable business goals. It says clients typically break even within three to six months, with profits increasing by an average of 17%. Similarly, Parallel Loop, with an outcome-based delivery model, has revealed securing eight enterprise AI clients within months of incorporation, with $2 million in booked revenue. But can large IT firms in India move with this model at scale? “Boutique AI firms are strongest in areas where incumbents would have to cut their own revenue to compete,” Leslie Joseph, VP and Principal Analyst, Forrester, tells AIM. India’s large IT services firms have traditionally relied on a mix of time-and-materials contracts, which bill for staff and hours deployed, and fixed-price contracts for an agreed scope of work. Neither model necessarily ties payment to business outcomes. For example, Infosys reported that fixed-price contracts accounted for 58% of its revenue in FY2026. Wipro, meanwhile, derived 45.6% of its revenue from fixed-price and volume-based contracts in FY2026, with the remaining 54.4% coming from time-and-materials contracts. However, outcome-based pricing is gaining ground as AI changes delivery economics, but remains a minority model. In August, Coforge, a Tier 2 Indian IT company, revealed that outcome-based contracts contributed 6-7% of its $2.5 billion revenue run rate. TCS has also reported a shift towards outcome-based commitments, particularly in agentic global business services “AI shrinks the billable work in these contracts, and clients now expect the savings passed back to them,” says Joseph. Boutique AI firms, by contrast, do not carry the same legacy contractual commitments. Instead, they make a compelling promise: to automate much of the routine work incumbents bill for. Joseph also explains that these firms can staff engagements with more senior engineers, positioning themselves as more AI-native than their larger competitors. Besides, outcome-based pricing is also viable for these firms in narrow workflows where results are easy to measure. “That model remains unproven for large transformation programs,” says Joseph. How to Create a Boutique AI Consulting Firm Many boutique AI companies are founded by engineers with direct experience developing and scaling software. They also use their personal experience and skills as a key selling point to attract clients. Verghese spent six years at Google, where he worked on Gemini projects within Workspace and helped develop techniques to reduce hallucinations and improve AI reliability. “Both (Verghese and his Co-Founder, Mridul Nagpal) of us loved working at Google, so we didn’t take this step lightly,” says Verghese. “Google set a particularly high bar for the reliability of these agents. In attempting to meet that bar, we adopted many techniques that we weren’t seeing people talk about publicly.” With Google's permission, Verghese and Nagpal began helping companies seek their advice and expertise. “We continued running the business while maintaining our roles at Google for over a year and a half. We promised ourselves that when the company hit $50k in MRR, we’d leave,” he says. “That happened in June of 2025, and by July, both of us had left.” Similarly, Jogani brings nearly two decades of experience and served as CTO of several tech companies across Asia before starting Parallel Loop. “My last company was listed on the Indonesian stock exchange, and we built the entire stack of it in four months,” says Jogani. However, one particularly interesting case is Cars24. The automotive logistics giant spent the last two years building and running AI agents for customer support, finance, vehicle inspections and other workflows. Rather than using AI for isolated tasks, Cars24 rebuilt parts of its business around AI-driven workflows, embedded in customer engagement, vehicle inspections, lending, engineering and internal operations. It said its AI voice agents handled 2.58 million conversation minutes in Q1 FY27, while chatbots processed more than 5,50,000 customer requests. AI-led systems conducted 80% of vehicle inspections in the quarter, up from 25% in the previous quarter. Its inspection platform reduced average inspection time from 45 minutes to 26 minutes. Similarly, AI-led loan disbursals rose to 65%, from 30% Having achieved these results internally, the company began exploring whether it could help more companies replicate its success. Thus emerged Deployment Inc. The Indian automotive sales giant has spun off its own AI consultancy firm after successfully deploying AI within its operations and now wants to replicate that success across the broader ecosystem. The company launched the firm in August, backed by an initial $5 million investment from Cars24 and co-founded by Jayesh Gupta, who led AI at Cars24, and Aayush Gupta, a former xAI engineer. “Cars24 is our first operating proof, and the experience there is what we are taking to other enterprises,” Jayesh tells AIM. “Implementing AI inside Cars24 taught us what happens after a successful demonstration: how users behave, where systems fail, how exceptions are handled and whether the economics hold.” For Deployment Inc, the initial focus is on founder-led, tech-enabled businesses, with meaningful operating scale. “We look for an important business problem, an executive sponsor, access to relevant systems and data, and an operating team that can work with us,” says Jayesh. He explains that a small team of forward-deployed engineers work closely with the customer’s operators and technology team, staying involved through discovery, building, integration, rollout and adoption. “We establish a baseline, put the workflow into production with appropriate controls, and measure whether it improves the business,” he adds. These firms can also
document fetched 2026-11-01 02:57