Analytics pods

A senior analytics and AI team, priced like a subscription.

Every pod includes an AI and analytics lead at a quarter allocation, works in your cloud and your repo, and ships on 21-day cycles after the first thread lands. Thirty days notice, any time, no exit fee. Analytics pods build the estate; AI pods build what stands on it. Start with one of the eight below, or build your own and we will tell you honestly whether the shape holds together.

Build your own pod Start with a Readiness Check
Start here

Analytics pods

The reporting estate itself: pipelines, models, dashboards people open, and the product analytics that settle arguments. This is where most engagements start, and where AI has to stand on later.

Foundation Pod

The pipelines and the warehouse that everything else stands on.

Right for you ifReports break on a Monday morning and nobody can say why.

1× AI & Analytics Lead · 0.25 FTE2× Data Engineer1× Analytics Engineer
4 months minimum First production pipeline in 14 days

What you get

  • Ingestion from your core systems, batch and streaming
  • A dbt project structured in layers with tests at every one
  • Orchestration, retries, alerting and on-call runbooks
  • Lineage you can show an auditor without preparing for it
  • CI/CD, so a change does not need a change freeze
  • A warehouse designed for your read patterns, not a vendor template
Enquire about this pod
Most asked for

Insight Pod

One set of numbers the whole company argues from, not about.

Right for you ifYou have data in four systems, three spreadsheets and one analyst drowning in requests.

1× AI & Analytics Lead · 0.25 FTE2× BI Engineer1× Analytics Engineer
3 months minimum Executive dashboard live in 21 days

What you get

  • A KPI framework signed off by the leadership team
  • A governed semantic model in Power BI, Fabric or Tableau
  • Executive, commercial and operational dashboards people open unprompted
  • Row-level security and a self-serve layer your teams extend themselves
  • Report performance tuned so people explore rather than raise a request
  • Metric definitions documented, version controlled and reconciled to finance
Enquire about this pod

Product Analytics Pod

Your product team stops arguing about whose number is right.

Right for you ifYou ship features and then debate whether they worked, because the events were named by whoever built them.

1× AI & Analytics Lead · 0.25 FTE1× Product Analyst1× BI Engineer1× Analytics Engineer
4 months minimum Funnel and cohort view live in 30 days

What you get

  • An event taxonomy and tracking plan your engineers will actually follow
  • Acquisition, activation, retention and revenue reporting
  • Funnel, cohort and segmentation analysis that survives a definition change
  • An experimentation framework with power calculations and readout templates
  • Identity stitching, so pre-login behaviour is not thrown away
  • A weekly product metrics review your team runs without us
Enquire about this pod

Copilot Pod

The business asks in plain English and gets your answer, not a plausible one.

Right for you ifCopilot is switched on, the answers are wrong often enough that nobody trusts it, and usage is falling.

1× AI & Analytics Lead · 0.25 FTE1× Analytics Engineer1× BI Engineer1× AI Engineer
3 months minimum Scored against your real questions inside 30 days

What you get

  • Semantic models tuned for natural language: synonyms, hierarchies, verified answers
  • A metric layer the assistant routes through instead of writing SQL from scratch
  • An eval set of the fifty questions your business actually asks, scored weekly
  • A refusal path — it says it does not know rather than inventing
  • Row-level security carried through into every AI answer
  • Adoption reporting: what gets asked, what gets answered, what fails
Enquire about this pod
Then

AI pods

What you build once the numbers are trustworthy: a context layer a model can read, decisions with guardrails on them, and agents that watch the business without making things up.

Context Pod

AI that answers with your numbers, not numbers it invented.

Right for you ifYou want to put a model in front of your data and you already suspect it will make things up.

1× AI & Analytics Lead · 0.25 FTE2× Analytics Engineer1× Data Engineer
3 months minimum First governed metric set exposed to an agent in 21 days

What you get

  • A dbt project with metrics defined once, tested, and versioned
  • Table, column and metric descriptions written for a model to read
  • Data contracts on the sources that actually matter
  • An MCP server exposing your metric layer rather than raw tables
  • Conformed dimensions, so two domains can be compared honestly
  • Retrieval over your documented business logic, not over a folder of PDFs
Enquire about this pod

Decision Pipeline Pod

A decision that runs itself, and can still be explained afterwards.

Right for you ifYou are making the same call thousands of times a day by hand, or with a rule set nobody remembers writing.

1× AI & Analytics Lead · 0.25 FTE1× Decision Scientist1× AI Engineer1× Data Engineer
4 months minimum One decision live end to end in 6 weeks

What you get

  • The decision mapped end to end: inputs, policy, model, override, outcome
  • A model in production, benchmarked against your current rules
  • Reproducible feature pipelines and a feature store where it earns one
  • Human-in-the-loop above a confidence threshold you set
  • Every decision logged with its inputs, its reason and its version
  • Champion and challenger running in parallel from day one
Enquire about this pod

Guardrail Pod

An AI system you would be comfortable putting in front of a regulator.

Right for you ifYou already have AI in the path and you cannot currently prove what it did, or why.

1× AI & Analytics Lead · 0.25 FTE1× AI Engineer1× AI Governance & Risk1× Analytics Engineer
4 months minimum Eval suite and tracing live in 21 days

What you get

  • An eval suite: golden sets, LLM-as-judge, regression on every prompt change
  • Grounding checks, so an unsupported answer fails instead of shipping
  • Input and output validation, PII redaction and refusal handling
  • Tracing per call: prompt, retrieved context, tokens, latency, cost, outcome
  • Model inventory, model cards and risk classification
  • EU AI Act Annex III readiness where you are in scope
Enquire about this pod

Monitoring Pod

The business watches itself, and only interrupts you when it matters.

Right for you ifNobody opens the dashboard until something has already gone wrong, and by then the quarter is written.

1× AI & Analytics Lead · 0.25 FTE1× AI Engineer1× Analytics Engineer1× Data Engineer
4 months minimum First KPI under autonomous watch in 21 days

What you get

  • An agent per KPI domain — revenue, marketing, product, risk, operations, cash
  • Baselines per series, so a threshold is learned rather than guessed at
  • Decomposition before escalation: the agent finds the driver, not just the move
  • A daily written brief and an alert path that respects your tolerance for noise
  • Every alert traced to the rows behind it, in one click
  • A weekly false-positive review that tightens the agents rather than muting them
Enquire about this pod
Pod builder

Build the pod you think you need

Add and remove specialists and watch the coverage change. The warnings underneath are the same ones we would raise on a call — an AI pod with the wrong shape fails quietly and expensively, and we would rather say so before you buy it.

Start from a pod, or from scratch
Specialists
Your pod
Headcount
—
Allocation
—
Delivery rhythm
—
Capability coverage
Send this pod to us

Coverage percentages are our own shorthand for how much of each capability area a given team shape can carry. They are a conversation starter, not a guarantee.

Start here

AI & Data Readiness Check

An honest answer to two questions: where AI would actually pay in your business, and whether your data can carry it yet.

Fixed fee 2 weeks
Book the Readiness Check
If you go on to run a pod with us within 60 days, the fee comes off your first invoice. The plan is yours either way.

Two weeks, and you own the output

  • Interviews with the people who use, and mistrust, the numbers
  • A review of your pipelines, dbt project, semantic models and reporting
  • Any AI already in the path, and the specific things it would fail on
  • Your EU AI Act exposure, if you have any, in plain language
  • A scored assessment across the five capability areas
  • A prioritised 90-day plan, with the first working thread specified
The people

Who is in a pod

Eight roles. Every pod is a combination of them, and every pod has the lead. Three of the eight did not exist in this shape three years ago.

AI & Analytics Lead

Sets the metric definitions everything else depends on, decides what should be automated and what should not, reviews every model, report and prompt, and is the one person you call.

  • KPI framework and metric definitions
  • Where AI pays, and where it will not
  • Architecture, model and eval review
  • Regulator, audit and board conversations

BI Engineer

Turns a warehouse into something the business actually opens. Semantic models, DAX, row-level security, and dashboards people trust enough to argue with in a meeting.

  • Power BI, Tableau and Fabric delivery
  • Semantic and dimensional modelling
  • Advanced SQL and DAX
  • Report performance and capacity tuning

Analytics Engineer

Owns the layer everything reads from — dashboards and AI alike. dbt models with a declared grain, metrics defined once and tested, and descriptions written for a machine as well as a person.

  • Conformed dimensional modelling: declared grain, one join path
  • Metric definitions, data contracts and tests
  • dbt project structure and CI
  • Documentation a person and an agent can both use

Data Engineer

Builds the plumbing everything above depends on. Batch and streaming ingestion, orchestration, retries, tests and a lineage you can put in front of an auditor.

  • Azure Databricks, Data Factory, Synapse
  • Snowflake and lakehouse modelling
  • Airflow orchestration and CI/CD
  • Data quality tests and monitoring

Product Analyst

Answers why the number moved. Owns the event taxonomy, the funnel, the cohort view and the experiment readout your product team argues over on a Monday.

  • Event taxonomy and tracking plans
  • Funnel, cohort and retention analysis
  • Experiment design, power analysis and readouts
  • Segmentation and lifetime value modelling

AI Engineer

Builds the system around the model: retrieval over your governed data, tool and MCP interfaces, agent orchestration, and the eval suite that tells you when a prompt change broke something.

  • Retrieval over a governed semantic layer
  • Agents, tool use and MCP servers
  • Eval suites and regression testing
  • Tracing, cost and latency observability

Decision Scientist

Owns the models that sit in the decision itself. Scoring, fraud, pricing and forecasting — benchmarked against a boring baseline, monitored for drift, and documented for a risk committee.

  • Scoring and propensity models
  • Fraud and anomaly detection
  • Forecasting and pricing models
  • Champion/challenger and drift monitoring

AI Governance & Risk

Makes the system defensible. Model inventory, model cards, eval evidence, human-in-the-loop thresholds, and the EU AI Act paperwork if you are carrying a high-risk use case.

  • AI inventory, model cards and risk classification
  • EU AI Act Annex III readiness
  • Eval evidence and sign-off trails
  • Bias, fairness and adverse-action testing

Not sure which pod you need?

That is what the Readiness Check is for. Two weeks, fixed fee, and you keep the plan whether or not you ever run a pod with us.

Book the Readiness Check Read how we work

Usually a reply the same working day · IST (UTC+5:30)