Invention Engine

Check the two hard filters first

• Stage. ACE is aimed at startups that have been operating for roughly six to twelve months and already have a product or an MVP, ideally with a few users. The accelerator says it typically takes teams from prototype or MVP through early traction and early scale, and will look at earlier teams only when the problem and the ambition are unusually strong.
• Location. You need to be able to move to Delhi NCR for most of the six months. This is stated as a condition of applying, not a preference, so settle it inside your team before you write anything.

Students are not excluded. The accelerator invites people still in college to apply, but only with a working product.

What to show in the application

• A product someone is already using. The clearest signal across every line of the eligibility list is a working build, not a deck. Name your users, even if there are only a handful.
• A specific customer. The first thing the programme works on with its teams is the Ideal Customer Profile and user segmentation. Arriving with a sharp, narrow definition of who buys and why puts you ahead of the first stage of the programme rather than at the start of it.
• Technical depth. The accelerator says it backs high-agency, technically strong founders, and it was set up by AI entrepreneurs. Be concrete about what you have built and what is genuinely hard about it.
• A global angle. Its advisory board exists because Indian AI startups selling into the US and Europe often misread Western buyer behaviour and enterprise sales. If you have any evidence of demand outside India, lead with it.

How selection works

The published process has four steps: an invitation for applications, one-to-one nurturing sessions with the accelerator's founders, a one-week in-house evaluation programme that ends in an offer, and then the cohort start.

Two things follow from that. The evaluation is a week of working together, not an interview, so expect to be judged on how you think and ship under pressure rather than on how you present. And the one-to-one sessions come before the evaluation, which means the relationship starts early — treat the first conversation as part of the assessment.

The accelerator's application page also refers to roadshows, which it says are mentioned in the application form. It does not publish where they are held or whether selection runs only through them, so ask when you apply.

Before you apply

• The accelerator invests in the companies it selects, and it does not publish whether it takes equity or on what terms. Ask before you commit, and get the instrument in writing.
• No fee is mentioned anywhere on the site.
• No closing date is published for the current cohort, so apply as soon as you are ready rather than waiting for a deadline.
• It backs AI and deeptech only. If your product is not core to either, this is not the right programme.
• General questions can go to the address published in the footer of the site, admin@inventionengine.in.

Invention Engine is a founder-led accelerator for early-stage AI and deeptech startups. It runs one programme, a six-month accelerator called ACE, and its published address is at 91 Springboard, Udyog Vihar, Sector 18, Gurugram, in the Delhi NCR region. The footer of its site names the operating entity as Big Bets Accelerator LLP.

The accelerator describes itself as a co-founding partner rather than a mentor panel. Its own words are that it embeds like a co-founding team and works side by side with entrepreneurs for six months, and that it was built by AI entrepreneurs rather than consultants or academics. Three people are listed as its founders: Varun Aggarwal, Shailendra Nath Jha and Shubham Bansal.

How ACE works

ACE runs for six months and is described as high-intensity and based in Delhi NCR. The accelerator invests in each selected company, gives more than 120 hours of one-to-one time with its own founders, and handles recruiting, finance and legal work so the team can stay on product and go-to-market.

The programme is laid out as six stages. It starts with backing high-agency, technically strong founders, then moves to nailing the Ideal Customer Profile and sharpening user segmentation, iterating the product through user feedback and hands-on product sprints, running tight feedback loops to get closer to product-market fit, shaping a fundraising and scale strategy with investor-ready narratives and models, and finally crafting a focused go-to-market strategy and testing real traction channels.

Cohort companies also get early access to a Global Customer Advisory Board. This is a group of senior operators based in the US, Europe and Singapore who help founders validate products, sharpen positioning and work out how to sell abroad. The site lists more than twenty of them, in product, engineering, sales and consulting roles at companies including Meta, Apple, Amazon, Microsoft, Google, AMD and McKinsey. Alongside it are an AI advisory group of researchers and practitioners, and a set of ACE mentors drawn from product leadership and founder backgrounds.

Two anonymised case studies on the ACE page show the kind of work involved: repositioning a niche edtech video tool as a creator-first AI video platform, and helping a horizontal voice AI company pick BFSI as its wedge, close its first large enterprise contracts and raise seed capital through the accelerator's investor syndicate.

Who can apply

Invention Engine states four things it looks for. The startup should have been operating for the last six to twelve months and have a product or an MVP, preferably with a few users. The team should be able to move to Delhi NCR for most of the six months. It is aimed at people who have worked in product-based companies and want to build in AI, and at anyone keen to build products from zero to one, including people still in college, provided they have a working product.

The accelerator says it typically works with startups from prototype or MVP stage through early traction and early scale, and that it stays open to exceptional teams at an earlier stage if the problem, approach and ambition are compelling. There is no stated incorporation requirement, no state or category restriction and no minimum funding bar. The only hard condition on the page is the relocation one.

The sector focus is narrow and explicit: AI and deeptech. The accelerator says plainly that it is not a scouting platform for funds.

How to apply

Applications run through an online form reached from the Apply for Cohort button on the home page and the ACE page. The published timeline has four steps: an invitation for applications, one-to-one nurturing sessions with the accelerator's founders, a one-week in-house evaluation programme ending in an offer, and then the cohort start. The current call on the site is for the 2026 cohort, with no closing date published.

Portfolio

The cohort page names Cheerio, an AI CPaaS platform for no-code messaging automation; Desible AI, an agentic voice AI platform for BFSI; SuperStars AI, a fantasy cricket platform built around creator-driven gameplay; Flowbee AI, an assistant for growing on LinkedIn; Rapture Innovation Labs, working on personal audio; Magicroll.ai, which transforms video using AI; and Callprep, which prepares sales teams before customer calls. The accelerator's news strip also features Sonic Lamb, a deeptech audio company that appeared on Shark Tank season four and was incubated in the 2024 cohort, and Magicroll.ai winning the AI, DeepTech and Cyber category at the Startup Maharathi Challenge during Startup Mahakumbh after the 2023 cohort. Callprep was selected for WTFunds. In March 2026, Indian startup press reported that four of its portfolio companies — Sonic Lamb, Cheerio AI, Desible.ai and Magicroll — had together closed a new funding round.

Invention Engine publishes no total for startups supported, no cohort size and no founding year, so the scale of the programme cannot be stated from its own pages.

Beyond the accelerator

Outside ACE, the team publishes free material for founders who are not in a cohort. There is a podcast featuring AI practitioners, a blog of articles on AI products and startup practice, recorded talks and presentations, and an India's State of AI Research report that measures India's output at the main AI research conferences and sets out recommendations for improving it. These sit on the Resources section of the site and need no application.

Invention Engine does not run a campus, labs or coworking space, and it claims no government or scheme recognition on any of its pages.