For Indian businesses · The BAIS method
Buying ChatGPT seats does not make a business AI Native. Seven stages do, in order, starting with the processes most Indian businesses never wrote down. Find out which stage you start from and what to fix first.
24 questions · 8 minutes · written for 14 business types
Why a roadmap, not a score
In owner-run Indian businesses, the gap is rarely tools. It is loose process: work that lives in the owner’s head, on personal WhatsApp and in Excel files on three laptops. AI cannot use knowledge that was never written down. So the useful question is not how AI Native you are. It is what to do next, and in what order.
Process, data and governance come first. A business cannot rise more than one stage above its weakest foundation, however many subscriptions it buys.
Every high claim is checked against something you could show a visitor. The roadmap starts from what is real, not from what the owner believes.
Each phase names three or four actions in your business's own language, with a target stage and the BAIS programme that supports it.
The seven stages
Stages 1 to 3 are about using AI. Stages 4 to 7 are about the business being built so AI can work inside it. Stages 4, 5 and 6 map to the BAIS method of Context, Loop and Judge.
Work runs on memory, paper registers and personal WhatsApp. The owner approves everything by phone.
What blocks the next stageNothing is written down, so nothing can be handed to a person or a machine.
A one-day session for owners and leaders on your own documents, and a written AI usage policy within two weeks.
Some staff use free ChatGPT on their own phones. Customer numbers and price lists go into free tools.
What blocks the next stageNo approved tools, no policy, no shared way of working. Results depend on the person.
A shadow-AI audit, a usage policy staff sign, and business accounts in place of personal ones.
A paid seat or two, marketing using AI for posts, and one person known as 'the AI guy'.
What blocks the next stageSOPs, product knowledge and brand voice are not written, so AI cannot be given context.
Four to six weeks: the 10 SOPs that matter, context packs for two functions, and 10 staff through the BAIS professional track.
A context pack per function: tone, catalogue, prices, policies. Everyone starts from the same templates.
What blocks the next stageAI is still used task by task. It is not inside the flow of work or connected to live data.
Eight to twelve weeks: three repeat workflows rebuilt with AI steps and review gates, connected to your systems, with time saved measured.
Repeat workflows run through an AI step with a review checklist. Time saved is measured.
What blocks the next stageOnly a few areas. Governance, training and measurement are not organisation-wide.
Roll out across functions, governance and DPDP-aligned data rules, training for every role, and internal AI champions.
Company accounts only, a monthly AI review, errors logged, and agents working inside systems with approval gates.
What blocks the next stageThe business is still designed the old way, with AI added on.
Ongoing partnership to redesign roles and services around AI, build agents, and place BAIS-trained associates.
Small teams run large operations. New services exist that were impossible before. Context is kept current like stock.
What blocks the next stageThe top stage. The work is to keep context current and stay ahead of competitors.
A retainer for new AI-led services, agent operations and a pipeline of BAIS-trained talent.
What we assess
Each dimension has three questions. Weights reflect how much a dimension holds a business back.
Whether the owner uses AI personally, who owns it, and whether it has a budget.
Whether repeat work is written down, done consistently and approved on record.
Whether core data and customer conversations live somewhere AI can reach.
Whether tone, offerings, prices and policies are packaged for AI to use.
How many staff use AI weekly, how they were trained, and how you hire for it.
Whether AI sits inside repeat workflows, connected to your systems, with review.
Policy, data handling under the DPDP Act, and how AI mistakes are caught.
Whether time saved and business results are measured and reported.
Scoring method · v1.1
The method draws on established maturity models (CMMI staged levels, the Nielsen Norman UX maturity stages, Gartner’s AI maturity model) and adds corrections for how self-assessments go wrong in owner-run businesses.
Each question gets its own screen. Answers describe observable behaviour, anchored to stages 1, 2.5, 4, 5.5 and 7. There is no “rate yourself 1 to 5”: owners overrate themselves on vague scales and rate accurately on concrete ones.
Any answer at stage 4 or above asks whether you could show proof to a visitor today. Unproven claims score one step lower. The roadmap reports both a claimed and an evidenced view; the gap between them is the say–do gap.
Respondents can skip, but a skip is read by its reason, not treated as a blank.
Six or more skips mark the roadmap as low confidence, with a guided diagnostic recommended before any spending.
Your starting stage is the lower of floor(weighted index) and floor(weakest foundation) + 1. If a foundation is holding you back, the roadmap names it as your first gate and puts it first in Phase 1.
The same questions are written for 14 business types, so a hotel’s 4.0 on data means the same as a foundry’s 4.0, and results can be compared across sectors.
What you get
Each dimension sits in one of four bands. Phase 1 gives the next action for each dimension from its current band, ranked by weight and gap, with foundations first. Each later phase moves every dimension up one band. Actions are written in your business’s own terms: a QSR is told to fix stock indents, a college its admission follow-ups.
Reach the next stage. Usually the first foundation: SOPs, one source of truth, a signed AI policy, and your first context pack.
Rebuild repeat workflows with AI steps and review checklists, connect them to your systems, and measure time saved.
Govern at scale, report returns in MIS, and redesign roles and services around AI.
Working with BAIS
The online assessment. One respondent, evidence discount applied. Gives a starting stage and a first-draft roadmap.
The owner, one manager and three staff answer separately, and we review the evidence in person. Gives a verified stage and a 90-day plan.
We sit with staff through three repeat workflows and time them. Gives a baseline, in hours and rupees, to measure every later step against.
Calibration: the weights and anchors are expert priors, not yet statistically validated. After 100 assessments we will check internal consistency per dimension (Cronbach’s α above 0.7), test the weights against outcomes such as time saved, and publish v2.
Answer 24 questions written for your type of business. Your roadmap is ready in about eight minutes.
