Trial-batch records and next-batch picks for small coatings and adhesive labs
TL;DR Recipewell is a lab notebook for small Indian formulators that also earns its keep. It records every trial batch and the tests run on it, then suggests which combinations to try next. Paint, adhesive and construction-chemical makers with 5 to 30 lab staff pay a monthly subscription per site.
A mid-size Indian formulator keeps its recipes in spreadsheets, a chemist's notebook and one senior person's memory. So when a customer asks for a lower-odour version, or a raw material goes out of stock, the lab starts from scratch and often repeats batches someone already ran two years ago.
The research points the same way. A 2025 review of machine learning for polymeric coatings names data scarcity as the main barrier to using these methods in practice[1]. Put simply, the models are there but the recorded data isn't. Separately, work on multi-objective formulation shows that statistical search methods cut the number of experiments needed when several properties must improve at once[2].
Recipewell has two halves that have to ship together. The first is plain record-keeping that a bench chemist will actually use - scan or type a batch sheet, log ingredients and quantities, attach test results and tag the customer or application. The second is suggestion. Once a product line has 30 or more logged batches with comparable tests, the software proposes a handful of next mixes aimed at a stated goal, such as holding gloss while cutting a solvent, and says which past batches back each suggestion.
The seed proposed discovering sustainable materials with AI. Discovering new chemistry needs a research lab and years of funding, which rules it out as a first business here. The same methods are useful one step down, where known ingredients get recombined, and that's where the paying customers are and where you can show a result in months rather than years.
The user is a formulation chemist or lab manager. The payer is the owner or technical director of a small or mid-size manufacturer, think paint, adhesive, sealant, construction-chemical and speciality-ink makers with their own lab. I found no public figure for what Indian SMEs spend on lab software, so a subscription of ₹30,000–1,00,000 a month per site is our assumption to test.
India's bio-industrial segment reached US$72.6 billion (about ₹7 lakh Cr) in 2023, with bio-based materials named among its growth areas[3], so more formulators are being asked to swap inputs. The BioE3 policy is also funding biomanufacturing centres expected to support over 250 startups and smaller firms[4], and that cohort starts with no legacy records at all and can adopt a system from day one.
There's no published figure for formulation R&D software spending by Indian SMEs. The honest signal is what vendors say about the tier above. One materials-informatics platform claims more than 500 R&D teams across 12 industries use it[5], and another describes one customer deployment with over 1,000 users at more than 35 sites[6]. Both are enterprise-scale, which tells you the category sells, not that your segment will.
One vendor sells a data platform with predictive models for materials R&D and claims most users cut failed experiments[5]. A US vendor covers R&D, quality and product lifecycle data, with coatings and paints among its listed industries[6]. A third serves chemicals and materials firms and claims five times faster development[7]. All three are built for large manufacturers. The gap is a stripped-down version an Indian SME can run without a data team.
| Difficulty | Moderate (38/100) |
| Starting capital | ₹50–90 L |
| Time to launch | 7–10 months |
The modelling is the small part. The real work is getting chemists to log batches in a form a model can use, which means putting up with messy entry, scanning old sheets and matching inconsistent test names. Expect to do the first customer's data clean-up by hand, and charge for it.
No sector licence applies. This is business software for manufacturers, so you only need the baseline.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't let AI sign off a formulation. A senior chemist decides what gets run, and suggestions built on thin data should say so.
Want to work out the AI layer for Recipewell? Talk to Yash →
I think this is worth a look if you've worked inside a formulation lab and can talk to a technical director as a peer. If you're a pure AI team then it's a poor fit, because the first year is data entry and trust-building. Prove one thing - that a real customer's logged batches produce a suggestion their senior chemist agrees is worth running.
Plan for roughly ₹50–90 L, mostly on product, data clean-up for the first customers and slow SME sales.
No sector licence. You need company registration, GST, a Shops and Establishments registration and DPDP compliance for staff contacts.
Yes, small paint and adhesive makers everywhere keep recipes in spreadsheets. Check your local data protection law and price for SME budgets.
The free Recipewell validation kit has the line-by-line capital breakdown, the effort scoring, the data-protection and contract checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for lab managers, and 5 assumptions to test. Download the kit →
Thinking about building Recipewell or something close to it? The make-or-break work is persuading chemists to log every trial batch in a form the software can learn from, and that's a habit problem more than a tech one. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.