DigitAI is Digitata's agentic AI platform. Not a chatbot pointed at a dashboard - a set of agents that work inside your environment, take multi-step action against your own systems, show their working, and stop at the gates you set. Analyst, on-call engineer and developer capacity, without the headcount.
A generative assistant produces text. An agent decomposes a goal into steps, executes them against real systems, checks its own output, and comes back with a result and the evidence behind it.
DigitAI agents query warehouses, read logs, correlate across services, write and test code, and produce finished deliverables. Each step is logged, each claim is traceable to its source, and each consequential action waits for the approval gate you configured.
Breaks an ambiguous business question into an ordered sequence of retrievals, joins, checks and computations.
Queries the warehouse, calls APIs, reads logs, runs tests, opens merge requests - with scoped, least-privilege credentials.
Reconciles figures against a second source, flags anomalies, and refuses to answer where the data can't support it.
Every number carries the query behind it. Every conclusion carries the trace from source to statement.
Each tier is independently deployable and independently priced. Most customers start with Analytics, add Operations once trust is established, and adopt Build when they want capacity rather than answers.
Ask, and get a verified answer. Plain-language questions across your warehouse, traced source to report, delivered into Slack or Teams.
The on-call engineer that never sleeps. Health checks, log reading and root-cause investigation - summaries instead of dashboard babysitting.
Development capacity on demand. Works in your codebase, tests before it ships, and opens merge requests behind a human approval gate.
Most analytics agents fail because they understand schemas but not semantics. DigitAI carries a business context layer: what a recharge is, why dormancy differs from churn, which revenue table is authoritative at month end.
Ask across the warehouse without SQL. Ambiguity is resolved by asking you a clarifying question, not by guessing.
Every figure is attributable - the query, the tables, the filters and the time window are attached to the answer.
"Why did recharge revenue drop last week?" returns a hypothesis with the segments, drivers and evidence behind it.
Branded Word, PDF or slide packs generated on a cadence and dropped into Slack or Teams - no analyst in the room.
Definitions are centrally curated, so ARPU means the same thing in every answer the agent gives.
Where the data cannot support a claim, the agent says so and names what is missing.
| What you ask | What DigitAI does | Tier |
|---|---|---|
| "Why did recharge revenue drop last week?" | Investigates across the data, attaches a root-cause hypothesis and shows its working | Analytics |
| "Which segments are drifting to dormant?" | Traced answer off your live behavioural profiles, with segments named and sized | Analytics |
| "Send me a weekly board pack on campaign ROI" | Branded Word or PDF deliverable into Slack or Teams, on schedule | Analytics |
| "Did the overnight subscriber ingest run clean?" | Reads logs, correlates errors across services, summarises in plain language | Operations |
| "Are the campaign sends and SMS gateways healthy?" | On-demand health check across services - up, flowing, nothing regressed | Operations |
| "Something broke overnight - what happened?" | Investigates the incident end to end, assembles evidence and names the cause | Operations |
| "Add a new bundle type to the recommender" | Branches, builds, tests on a dev server, then raises a governed merge request | Build |
Operations agents watch the estate continuously and investigate when something moves - so your team reads a summary instead of watching dashboards.

Build agents work in your repository the way a disciplined engineer does: on a branch, with tests, against a dev environment, ending in a merge request a human reviews.
Work starts from a written requirement - a ticket, a change request or a plain-language brief - with acceptance criteria agreed up front.
The agent works on its own branch in an isolated session. No access to production data or production credentials.
Code is written to your conventions, unit and integration tests are run, and the change is exercised on a development server before it is offered.
A human reviews a normal MR - diff, test output, and the agent's own account of what it changed and why.
After merge, the Operations tier watches the affected services and reports regressions against the pre-change baseline.
Agentic systems fail on governance long before they fail on capability. DigitAI ships with the control layer built in - the same framework Digitata applies to its own regulated deployments.
Every agent run is sandboxed. No shared state between customers, environments or sessions.
Credentials are scoped per task and per tier. Analytics is read-only. Nothing gets standing production write access.
Consequential actions - sends, deployments, schema and pricing changes - stop for named human approval.
Every prompt, tool call, query and output is logged and replayable for internal audit and regulators.
Agents earn scope. Each capability starts supervised and widens only on demonstrated accuracy.
AI management system practice aligned to ISO/IEC 42001, with model, data and change governance documented.
DigitAI is not a general model pointed at an operator. It carries two decades of Digitata data science across 20+ mobile markets: prepaid behaviour, recharge cycles, USSD constraints, corridor and regulator nuance.
No per-seat licence fees. Adopt Analytics, Operations or Build independently and scale between them.
Whole teams - commercial, operations, engineering - across every brand you run.
Continuous tuning, quality monitoring and quarterly roadmap reviews are part of the service, not an add-on.
Lean commercial and engineering teams carrying more scope than headcount - MNOs, MVNOs, FTTH operators, banks and money transfer businesses running Digitata platforms or their own.
DigitAI sits above your operational platforms - including Vaitom and Digitata Behavioural Banking - as the reasoning and action layer across them.
We'll scope a first tier against your real environment and show you the agent working on your own data - with the governance gates in place from day one.