It showed us the mirror. We had no system of record.
Gartner’s hype cycle maps how organizations move through any new technology – from inflated expectations, through disillusionment, towards eventual productivity.
Enterprise AI fits the bill perfectly.
I’ve witnessed a few such cycles – here’s one that took us through the full emotional range.
The Promise
Recruitment, averaging 10k closed positions per year, was to become fully automatic with very little human involvement. Automation solutions for isolated hiring activities like CV screening, video interviews, interview scheduling, etc. had been in the market for a few years.
Specialized AI agents provided the missing pieces of the puzzle to stitch together an an automaton that worked from Job requisition to Offer rollout.
Very doable and realistic, with a flurry of off-the-shelf solutions, and dozens of consultants to help with design and implementation.
RFP was closed and partner selection due in a week.
Ribbon Cutting: Month 1
A cross-functional Steering Committee formed, liaising with the Automation Advisory Board. The project was christened around a celestial object – and kicked off by the top boss.
Savings and productivity enhancements were compelling, and so simple that they could be modelled within the mobile view of a spreadsheet. The final super-agent was to handle the workload of 150+ recruiters. At the eventual pinnacle, a single administrator could run the entire machine through just a configuration panel and tweak it to business needs.
The timeline was liberally set to 9 months for the transformation with clear milestones and a phased Gantt chart that would make the Project Management Institute proud.
Agents would coordinate and aggregate across three HR platforms (Recruitment cloud, employee lifecycle management, application tracking system), MS suite for email and calendaring, recorded interviews on cloud, and the occasional unstructured data – cakewalk for AI.
First Cracks Appear: Month 3
External consultants, now based in the office full time, had been racking their brains to unravel the organizations within organizations, scoping scenarios where edge cases did not feel like exceptions, but norms – the tail seemed to be wagging the dog.
Implication – processes are not adequately documented and completing that would add a few weeks to the timeline. So far, so good.
However, it was turning out to be impossible to set a baseline. Recruiters, who had been consulted for the first time in years, did not agree on provenance of HR platforms. Cynics came out of the woodwork to claim the platforms were just approximations.
Many recruiters conducted their workflow offline, since the system gave them the liberty of logging in the candidate after the full cycle was complete and offer rolled out.
The odds of a candidate being in the Recruitment Management platform, though not 100%, were still better than those of Schrödinger’s cat being alive.
Revelations and Disillusionment: Month 4
Recruitment dashboards, built on top of the uncertain HR management tools, looked greener than the Amazon rain forests, while business stakeholders continued to flag delays and hiring quality.
Some current employees in the system seemed to have materialized without any trail, or even a trace, in the Recruitment platform.
There was truly no system of record, and it felt like we had been flying blind all these years. But somehow, things worked fine.
Existential Questions: Month 5
Glasses come off, heads tilt, shoulders sag, loud exhales
How did we miss this?
Who is responsible? Which heads should roll?
Did we even pick the right problem to solve?
How much did we pay the vendor already?
Are they even the right vendor for the exercise?
We need a war room.
Let’s fire the vendor.
AI has already evolved so much since we started. Let’s re-evaluate.
Silence and Contemplation: Month 6
The static of reflection and contemplation, having burned the hands once, and worrying about the second attempt.
Debriefs and huddles to decide the course forward while the questions from the board still loomed.
“What is our AI strategy?”
“What did we really achieve this year?”
Some consultants call this phase the ‘Valley of Death’ because not everyone makes it to the other side, and those that do, rarely come unscathed.
Epilogue
The agents never even had a chance to fail. The project collapsed under the weight of the data it was supposed to ingest—a system of record that existed only in our spreadsheets and imaginations.
Until we fix the plumbing, we’re just buying a more expensive faucet.
The board still wants answers. A new vendor is being evaluated. The Gantt chart is already being rebuilt.
Array
(
[0] => WP_Term Object
(
[term_id] => 26
[name] => POV
[slug] => pov
[term_group] => 0
[term_taxonomy_id] => 26
[taxonomy] => post_tag
[description] =>
[parent] => 0
[count] => 4
[filter] => raw
)
)