Two ways to buy. Engage the transformation practice, one workflow at a time, starting with a fixed-price scan. Or buy an agent from the workforce outright. Model inference, third-party API fees, and cloud storage pass through at cost with no markup.
For a personalised range based on your actual workflow, the estimator takes about two minutes and gives you a band you can put in a board memo.
The practice engagement, one workflow at a time. Start with a fixed-price scan, build the lighthouse, then run the fleet.
Two weeks. We map the workflow end to end (handoffs, decisions, exceptions), audit your context and data, set a governance baseline, and model the ROI per workflow.
Credited in full against a Lighthouse Workflow within 60 days.
One workflow redesigned, staffed with agents, grounded in deterministic context, and instrumented. First working version live in two weeks, hardened over six to ten.
The performance guarantee holds: miss the bar, we work free until it is met.
We run the fleet in production: monitoring, evals, guardian reviews, kill-switches, quarterly retrains, and a cost-per-outcome dashboard your CFO can read.
Priced to the fleet and the outcome, base plus optional bonus.
Prefer to buy an agent outright? Six roles, each priced differently because each does a different job.
The questions every buyer asks before the first call.
It starts small on purpose. A fixed-price Agentic Readiness Scan is $7,500 and takes about two weeks: we map the workflow, audit the context and data, set a governance baseline, and model the ROI. The scan fee is credited in full against a Lighthouse Workflow build (typically $35,000 to $120,000) if you go ahead within 60 days. Running the fleet afterwards is an Agentic Operations retainer from $4,000 per month. You never buy a multi-year programme up front. You buy one workflow at a time.
On Lighthouse Workflow and Agentic Operations, yes. We set a written baseline for the metric that matters (cost per case, deflection rate, cycle time), charge a base fee, and add a bonus above the baseline. We do not do pure outcome-only pricing, because the base fee keeps the incentives honest on both sides, and the performance guarantee already means we eat the downside if the agreed bar is missed.
Ranges vary by agent type and scope. Ready-made agents are $99 per week per role. Sales agents start at $299 per week. Second Brain knowledge agents start at $1,200 per month for teams up to 100 people. Predictive Legal and Server agents are scoped to enterprise needs and quoted per corpus size or fleet. The price estimator above gives you a personalised range in about two minutes.
Most agents ship working drafts or live responses inside two weeks. The first week covers intake, retrieval setup, and initial fine-tuning on your material. Week two is evaluation, human review calibration, and handoff. Complex multi-system integrations or regulated deployments such as legal or HR take three to four weeks.
Implementation, fine-tuning on your material, integration with your systems, evaluation harness, and the first quarter of maintenance. Model inference costs and third-party API fees such as WhatsApp, SMS, Stripe, or data enrichment are passed through at cost. There is no hidden markup on infrastructure.
Yes. Most agents can be piloted on a single workflow for four weeks before you commit to a full deployment. The pilot is scoped, priced, and has a written success criterion. If the agent does not hit the criterion, we work at no additional cost until it does.
It means the cost reaches you at the rate we pay the provider with no Perform Digital markup. WhatsApp template messages, SMS carrier fees, Stripe processing, model inference tokens on usage-billed models, and cloud storage are all pass-through. You see the exact provider invoice.
Weekly-priced agents such as Ready-made and Sales have no minimum beyond the pilot period and cancel any week. Monthly agents such as Second Brain run month-to-month after the pilot. Enterprise agents such as Legal and Server are annual with a 90-day break clause. Nothing auto-renews without written notice.
A senior AI engineer costs $180,000 to $260,000 per year in the US, plus equity and benefits, with a three to six month ramp time. Our agents are domain-fine-tuned products with evaluation harnesses, compliance controls, and maintenance included. You get a working agent in two weeks for a fraction of the annual engineering cost.
We write a success criterion before we start. If the deployed agent does not meet it at the end of the pilot, we continue working without additional charge until it does.