14%
more customer issues resolved per hour by support staff given an AI assistant, measured across 5,179 agents and 3 million chats.
Brynjolfsson, Li & Raymond, Generative AI at Work, NBER Working Paper 31161. nber.org- Home
- AI Agents
Coming soon
AI agents for the work that doesn't scale
A second service line in development: AI agents that take repetitive operational work off your team, so the people you employ spend their time on the work that grows the business.
In development — not yet available
Under construction
This page is a preview, not a product page
Nothing here is for sale yet, and there's no pricing because we haven't finished scoping it. Below is what we're building and the independent evidence behind it. Tell us what you need — early clients shape what ships first.
Register your interestThe short version
What an AI agent actually is
An AI agent is software that carries out a multi-step task on its own — reading a request, working out the steps, using the tools needed to do them, and reporting back. A chatbot answers a question; an agent handles the errand.
That suits the work that quietly eats a small team's week: qualifying leads, chasing reviews, assembling reports, triaging an inbox. It does not suit judgement calls — a limit we take seriously rather than gloss over.
The evidence
What controlled studies actually found
Every figure below comes from a named, published study — not a vendor case study. Full sources at the bottom of this page.
34%
gain for the least experienced workers in that study, while the most experienced saw minimal change. It raised the floor, not the ceiling.
Brynjolfsson, Li & Raymond, Generative AI at Work, NBER Working Paper 31161. nber.org25.1%
faster task completion among 758 BCG consultants, who also completed 12.2% more tasks at roughly 40% higher rated quality.
Dell'Acqua et al., Navigating the Jagged Technological Frontier, Harvard Business School Working Paper 24-013. ssrn.com9%
less time per customer chat with no measurable drop in satisfaction scores. Speed did not cost quality.
Brynjolfsson, Li & Raymond, Generative AI at Work, NBER Working Paper 31161. nber.org62%
of organisations are at least experimenting with AI agents, but only 23% are scaling one anywhere. The market is still early.
McKinsey & Company, The state of AI, 2025 survey. mckinsey.com10%
is the ceiling in any single business function. Being early here is still a real advantage.
McKinsey & Company, The state of AI, 2025 survey. mckinsey.comThe other half
Where the same research says agents fail
We'd rather lose a sale than oversell this. Two findings shape the design.
Outside its competence, AI makes people worse
The same BCG experiment included one task deliberately placed outside the model's capability. On it, consultants using AI were 19 percentage points less likely to reach the right answer. The researchers called that uneven boundary the "jagged frontier". So we scope agents to narrow jobs with a verifiable output, rather than pointing one at a whole operation.
The most-cited success story was partly walked back
Klarna is the deployment everyone quotes: in February 2024 its AI assistant handled two-thirds of service chats, the work of some 700 agents. Less quoted is May 2025, when it began rehiring humans after complaints about generic answers. It settled on triage — agents take routine volume, people take disputes and anything with money or trust at stake. That is the model we're designing toward.
Planned
Where we're starting
The agents we're scoping first, chosen because they sit close to the marketing work we already do. Nothing here has shipped.
Inbound lead qualification
Scores every enquiry against your criteria, drafts a first reply, and pushes the ones worth your time to the top.
Review generation & response
Follows up at the right moment and drafts on-brand replies, escalating negative reviews to a human before anything is published.
Reporting & anomaly alerts
Assembles your monthly report from source data, and flags ranking drops the week they happen instead of the month after.
Content operations
Turns a keyword plan into briefs and drafts, with a human editor approving everything before it reaches your site.
Booking & follow-up
Handles scheduling back-and-forth, sends reminders, and re-engages enquiries that went quiet.
Early access
Be first in the queue
Leave your email and we'll be in touch when early access opens. Tell us what eats your team's week — the first agents we finish will be the ones early clients actually need.
No newsletter, no drip sequence. One email when there's something real to show you.
References
Sources
- Brynjolfsson, E., Li, D. & Raymond, L. Generative AI at Work. NBER Working Paper No. 31161 (2023); published in the Quarterly Journal of Economics. Study of 5,179 customer support agents across roughly 3 million chats. nber.org/papers/w31161
- Dell'Acqua, F. et al. Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Harvard Business School Technology & Operations Management Unit Working Paper No. 24-013 (2023), conducted with Boston Consulting Group across 758 consultants. papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321
- McKinsey & Company, QuantumBlack. The state of AI. Global survey, 2025 edition. Figures cited: 62% of organisations at least experimenting with AI agents, 23% scaling an agentic system, no more than 10% scaling within any single function. mckinsey.com
- Klarna. AI assistant performance announced by company press release, February 2024; reversal and rehiring of human customer service agents reported May 2025 following comments by CEO Sebastian Siemiatkowski.
Figures are reproduced as published. They come from specific companies and task types and are not a forecast of any individual result. Photography: Unsplash License.
Available today
In the meantime, let's get you ranking
AI agents are coming. Google Maps SEO, AI SEO and Local Services Ads are ready now.
Book Your Free Consultation