AI implementation costs about $10,000 to $40,000 for a first pilot on one use case, according to one consultancy's published phases, and $150,000 to $500,000 for a first year of AI at a company of 100 to 1,000 employees, according to another firm's illustrative budget. Ecommerce use cases can start lower: one retail AI firm publishes $4,000 to $10,000 for a single use case. The development fee is only part of the cost. Implementation also pays for data cleanup, software and model fees, training the team, governance and a contingency, and moving from pilot to production costs several times the pilot. This page sets out published prices at each stage, the costs a development quote leaves out, a 90-day pilot budget for a DTC brand, and what scaling up adds. Brand Media Works implements AI workflow automation for ecommerce brands; the prices below are what other firms publish about their own work.
What does each stage of AI implementation cost?
Most firms sell implementation in stages: an assessment, a proof of concept or pilot on one use case, then a production build. The table sets out published prices from firms in the Brand Media Works competitor register, with firms that publish ecommerce and retail prices listed first. Each firm is described by the kind of work it does.
| Firm type | Assessment or discovery | Proof of concept or pilot | Production |
|---|---|---|---|
| AI automation and agentic AI development firm with a retail practice | Free scoping call | One retail use case, such as a shopping assistant, recommendations, dynamic pricing or demand forecasting: $4,000 to $10,000 over four to ten weeks, with 90 days of post-launch support | Full retail AI platform: $16,000 to $24,000 over 16 to 24 weeks |
| AI development studio with a retail and ecommerce practice (US and India) | $5,000, credited against the build | $30,000 fixed price, six weeks to a production deployment | |
| AI automation and custom AI agent firm | $999 assessment; readiness and governance sprint from $7,500 | Defined automation build from $8,000 | |
| AI consultancy and software development firm (Netherlands) | €2,500 audit and strategy | €20,000 proof of concept | From €50,000 |
| AI strategy and implementation consultancy | Discovery and pilot together: $25,000 to $75,000 over three to four months | Pilot on one use case: $10,000 to $40,000 over 60 to 90 days | Expansion: $40,000 to $100,000 over 6 to 12 months; organization-wide: $100,000 to $500,000 or more over 12 to 24 months |
| AI development and managed AI services firm (US) | Not published | Fixed-scope pilot: $25,000 to $50,000 over four to six weeks | Not published |
| Specialist AI and LLM consultancy (UK) | Not published | Proof of concept: £25,000 over three weeks; pilot with real users: £75,000 over eight weeks | £200,000 over 16 weeks, plus £40,000 for the first year of operations |
| Enterprise AI software vendor with a strategy practice | Not published | $15,000 to $250,000, most commonly $25,000 to $200,000, over 6 to 12 weeks | Departmental rollout: $150,000 to $1.5 million over 6 to 12 months |
| Founder-led AI consulting business | Readiness assessment: $2,500 to $8,000 | Pilot on one use case: $10,000 to $25,000 | Done-for-you implementation: $25,000 to $60,000 |
The cheapest published prices are for one use case built on the brand's existing tools with clean data. Prices rise with the number of systems the AI has to connect to, the state of the data, how much the AI may do without a person checking it, and where the team sits.
What is the difference between AI development cost and AI implementation cost?
AI development cost is the fee for building the software: the model integration, prompts, connections to other systems and testing. AI implementation cost is everything it takes to get that software used in the business. One firm's illustrative first-year budgets show how much sits outside the build:
| Company size | First-year total | What it includes |
|---|---|---|
| Under 100 employees | $30,000 to $80,000 | AI software tools $12,000 to $30,000; light integration and setup $8,000 to $20,000; data cleanup $3,000 to $10,000; training $3,000 to $8,000; contingency of about 25%, $6,000 to $16,000 |
| 100 to 1,000 employees | $150,000 to $500,000 | Platform and licenses $40,000 to $120,000; implementation and consulting $40,000 to $120,000; data readiness $25,000 to $80,000; infrastructure $15,000 to $60,000; governance $10,000 to $40,000; training and change management $15,000 to $40,000; contingency of about 30%, $35,000 to $120,000 |
For a company of 100 to 1,000 employees, the implementation and consulting line is about a quarter of the first-year total in that budget. The rest is software, data, infrastructure, governance, people and contingency. A brand that budgets only for the development quote will run out of money before the AI is in use.
Data readiness is the line that grows fastest: the same firm puts it at 8 to 12% of an enterprise AI budget, rising to 35% when the data is messy. Gartner predicts that through 2026 organizations will abandon 60% of AI projects unsupported by AI-ready data, and a Gartner survey found 63% of organizations either do not have or are unsure if they have the right data management practices for AI. For an ecommerce brand, that usually means order, customer, product and ad data that disagree between Shopify, the ad platforms and the email platform. How much marketing data engineering costs sets out what fixing that costs.
What should a 90-day AI pilot cost a DTC brand?
A pilot tests one use case with real users and real data, and ends with a decision to scale it, change it or stop. The table builds a 90-day pilot budget from the published prices above. Model and software subscription fees are left out, because many pilots run on tools the brand already pays for.
| Line | Weeks | Low | High | Where the figure comes from |
|---|---|---|---|---|
| Assessment and scoping | 1 to 2 | $999 | $7,500 | One automation firm's published assessment and readiness sprint |
| Pilot build on one use case | 3 to 10 | $4,000 | $40,000 | One retail AI firm's lowest use-case price; one consultancy's highest 60 to 90 day pilot price |
| Data cleanup | 1 to 6 | $3,000 | $10,000 | One firm's first-year budget for a company under 100 employees |
| Training and adoption | 8 to 12 | $3,000 | $8,000 | Same budget |
| Contingency at 25% | $2,750 | $16,375 | Same budget | |
| Total | 13 | $13,749 | $81,875 |
The range matches what one consultancy publishes for discovery and a pilot together: $25,000 to $75,000 over three to four months. Where a brand lands depends mostly on the build line: a retail use case on standard tools with an ecommerce specialist sits near the low end, while a pilot that connects several systems or needs custom data work sits near the top.
The weeks matter as much as the money:
- Weeks 1 to 2: pick one use case and write down the baseline. Choose a job that repeats daily and has a number attached, such as tickets per agent, time to publish a product listing or repeat purchase rate, and record today's figure.
- Weeks 1 to 6: get the data ready. Connect only the sources the use case needs and agree the definitions.
- Weeks 3 to 10: build and test. Start with a small group of users and a person checking the output.
- Weeks 11 to 13: measure and decide. Compare against the baseline and decide whether to scale, change or stop.
What does scaling an AI pilot to production cost?
More than most budgets allow for. One firm says production scale-up usually requires 3 to 8 times the pilot budget, so a $40,000 pilot implies $120,000 to $320,000 to put it into production across the business (our arithmetic). A UK consultancy prices a first production build at £200,000 over 16 weeks after a £75,000 pilot, and says that below £100,000 a buyer is getting a pilot, not production. Another consultancy publishes $40,000 to $100,000 to expand a proven pilot to more departments or use cases over 6 to 12 months.
Running costs come on top. Two firms put annual operations at 15 to 25% of the build cost, covering monitoring, model and prompt updates, and fixing integrations when other systems change. How much AI agent development costs sets out monthly running costs for a single agent, including model fees.
When should an AI implementation pay for itself?
One development firm says returns typically show up in 3 to 12 months, through productivity gains or lower costs. Another consultant works through an example: a $15,000 two-month build that settles at $6,500 a month of net value pays back its fee in month four. A pilot that cannot show a clear move against the baseline by week 13 is unlikely to pay back at production scale, which is why the baseline is written down before the build starts.
What should a brand ask before signing an AI implementation contract?
- What does the pilot price include? One named use case, the data sources it uses, the users who test it and the number it has to move.
- What is outside the quote? Data cleanup, software and model fees, training and contingency, each with an estimate.
- What happens after the pilot? A written estimate for production and for annual running costs before the pilot starts, so the decision at week 13 is not a surprise.
- Is the discovery fee credited? Some firms credit the assessment against the build.
- Who owns the result? Code, prompts, workflows and data pipelines should belong to the brand.
- Is support included after launch? At least one retail AI firm includes 90 days of post-launch support in its prices.
For most ecommerce brands the best first pilot automates one daily workflow in support, merchandising or marketing reporting, on data that is already in reasonable shape. AI workflow automation sets out how Brand Media Works scopes and runs this work. Published prices for every kind of AI consulting work are in AI consultant cost in 2026.
Frequently asked questions
How much does AI implementation cost for a small business?
One firm's illustrative first-year budget for a company under 100 employees is $30,000 to $80,000, including software tools, light integration, data cleanup, training and a contingency. A single pilot can cost less: published pilot prices for general business use cases start at about $10,000, and one retail AI firm prices a single ecommerce use case from $4,000.
How long does AI implementation take?
Published pilot timelines run from three weeks for a proof of concept to 60 to 90 days for a pilot with real users. Production builds are published at four to six months for a first capability and 6 to 12 months for a departmental rollout.
Why do AI pilots fail to reach production?
Usually because the data was not ready, the pilot had no agreed measure of success, or nobody budgeted for production. Gartner predicts that organizations will abandon 60% of AI projects unsupported by AI-ready data through 2026, and one firm puts production at 3 to 8 times the pilot budget.