everybody. our Thank afternoon, Good you joining for call.
a probably saw you today, we have in announcement earlier As our to you. of share exciting news quite lot with
from programs AI, selected received last weeks, technology the largest we to behind the have for been ChatGPT. that provide of X innovation global the Over data couple we in confirmation verbal generative X companies technology engineering of their
One the an of one a these customer. customer existing and is new companies other will be
and likely of on to reached them technology potentially accomplishments terms X Innodata. third that services master new us choose has We also a a a another company, global and indicated the agreement with these companies, believe are of largest addition, agreement. we just for transformative customer are In they have
be required For work build engineering these expect companies, a to we to generative of data performing range cutting-edge AI.
to by used instruction this models, learning, the to to expect providing instructions, creating We teach data that sets process train sets follow providing training reinforcement potentially data a that working models which model and the are red include values multiple complex performance we in and and use providing cases languages evaluation. several data across human align models This and modalities. teaming with involves potentially
positioning able of the both out landscape largest operate the our build and believe be we impact of I last, to this change, on how unleash get potential we believe fifth worldwide; and results forward and tremendous will tremendous view businesses second, generative year financial and traction this how as drivers fourth, call are will their to advantage been to rapid growth how take AI companies they beyond; generative the cadence these change; thinking and on today's going Innodata the forward; QX first: focusing third, have and guidance. we capabilities wins, seek the AI of Our tech we with why our from we're about industry
advantages, consciousness we AI Internet potential, the believe be we as with servers, not place into the provide. collective the XXXX, could that it believe productivity cetera, customer were will massive tremendous world for coming in Internet. in company the transformative most provide people the late develop. When were in technology have the because and will to as logistics, as and yet innovative Virtually believe IPO ad been even to but incorporate broadband, technology of than busted in et more will resulting the capabilities competitive. will to We generative like capital experience in monetization basic be were Internet's the revenues every businesses this the that infrastructure, And well see subsequently Netscape's destruction consumers slow
and much by case contrast, over is business adoption will valuable and cycle AI, of how we But in most of Of several believe realized. time, benefits created course, companies conducted. generative quickly was that the potentially when the more it built the infrastructure adoption world be in the of revolutionized
This is motor an among this XX% The engine MIT using the sector. AI in may the massive do productivity driven service increase expect industrial opinion, industrial for productivity. in the adoption referenced will and provide. of their using sector processes. We our productivity revolution leap for electric human ChatGPT adoption the of in paper what increases, did level by will versus to generative be the legacy steam generative by AI Recent workers likely
that entrants, is we AI enter amounts in that opportunity the foundation with are like announced believe AI and to raising build to are of race our as companies a immediate today large Anthropic, as generative capital cap companies, race. the Character We massive new tech mega as well
use These very resulting for and implementation We their proven. the is from fear potentially productivity companies the providing Moreover, much structures readily of of and cases. be may seem render available specific applications licensing their yesterday. do plan and and parties relatively rapid adoption the in cost them technology potentially not foundational AI to that the higher may that rapid, believe these technology incorporate of because more needed if on experience is its say build customer which this so primarily ecosystem on generating focused companies capabilities, uncompetitive enhancing niches to than these nothing they the likely Internet will for application because infrastructure generative and will layer third revenues AI they're to the by that own licensees
providing end solutions that second label services end their fine-tune opportunity to is basis for of provide to technology foundation with build a to models versions their do on engineering these these to foundation white their on to customers own these models. We a side-by-side the companies, data large the or The opportunity help customers basis. Innodata both work on services see and
progress front companies hope to report We doing are this next call. with by in have to X of active on that. discussions just these about our We
a viable we businesses data current that believe generative own The their based as We evolution the have increasingly We on attractive opportunity models. build making. X to large becoming is predictions become see help AI are will this increasingly third of sets. this to large businesses large likely unique likely path proprietary
increasingly open-source high-performing commercially that First, available. become generative models usable AI will
techniques effective will more marginal systems. fine-tuning progressively other GPUs that the for reduce, third, innovations and And compute thanks significantly over cost will, Second, to time, of that become model in available.
people to We today. master is data of the in leading context their agreement Their much enable believe, signing we example, their that with to LLMs quote data, to with query bank are scientists of people We're services year for this. XX evidence this team that the deployed already investment their vision seeing chat a to a databases close way earlier data.
of areas the building in and building model, main There are of of costs the X the costs engineering. essentially cost compute the generative data AI; of cost
engineering, time, less see costs, As come data compute investment. costs expensive over businesses as models that we cycles commoditize, down fine-tuning high a ROI the and will remaining large become expect
LLM become represents. will key and testing use data potential unlocking address will believe to it that cases. data required, including be their for advantage economic curation, We engineering for become creation, and them ambitious that the the preprocessing progressively collection, expect we competitive data Moreover, proprietary to will the more sophisticated that training work
we new point, believe We the are in rapid applications predict with enviable able This transitions applicable focused companies arms us to as gain did our has we AI generative sometimes it been nor not would why better burst system, smart. history long put quickly be that has, third are building an our for AI, scene the believe position. onto which of say other traction this to of but Innodata I as lucky such to directly architects is capabilities than leading it now on would has to race. generative we
placed sciences, civic thousands scalable We matter diverse the areas, pharma, and in around agriculture, world. domain experts haven't legal material biology, including math, with of expertise
high-quality a have consistently creating sets subjectivity. reputation proven areas in We complex subject for data with right
XX-plus have global a languages. work in reach, us We to enabling
real in the AI data ingest and learning deployment. to have we around have have acumen zero-shot the technical time. model we annotate to past auto and X quality years And training with digest platforms and developed to track developed We metrics over flexible data
starting to the made believe and be XXXX for generative tailor-made technology AI. We turned choices we out in back XXXX
For AI. example, we was chose generative time, encoder-based an At transformer the supported obvious that architectures choice. hardly this
models had GPT-X our Moreover, production we experience the first us today's our versus, environments essential is for example, are parameters, that into real-world years settled on reported XX X million parameters deploy safe X trillion value While to ways to ago. only and is what models taught how have architecture in that creating.
luck. not So all maybe it was
over several we and been AI platforms. the In workflows generative have classical addition, operations, past customer into integrating years, and real-world both
foundation developed recently, engineering deploy developed and overcoming and fine and challenges. building and models. LLMs important have and prompt We recipes prompt More deep in expertise trust chaining, tunable for required technologies configuring have skills to we safety
on a and customers going both we businesses experience as we implementation repurposable largest As invaluable. we plan with we to work hope to, basis, will be the side our our alone on large with the and believe as
customer of speak and our Fourth, These I us X they could give wins a the customers notified potential and today broadly these up win accomplishments how indication of -- the we in we past want show to magnitude over framework the and positive announced days. for a XX believe results. to
our of is global existing earlier, the mentioned customer, One of an order process the place I repeating, in but we customer for list new. of are tech the are X worth it's remaining win. putting in existing top X X customers the each these As With a while new work firms. the is in
place formal X the in new finalizing putting and customers, are With now initial scope. the we in agreements process of
cut is million it not. that Moreover, which $X.X go, that that large. be where approximately maturity. illustrate, new revenues chance To for happens, the purchase deals While are enable worked we that will quarterly to order work forward. customers now will potentially an vision The there are the new this scope. has of their that draft being get could inked initial confident might contracts They also one add scope customers' agreements easily that will are supplementing started, as move are them initial does indicated it in we million of fully it agreements realized, our but at us until be us always result on if shared believe requests, the the with $XX framework at of we for to
the intended Once results. in Again, process an but be resources, we and I that specifications we're specification subject up custom-configuring the them detailed This locked months. required will once customers emphasize and run this over signed, technology engagements for infrastructure. have data over pilots create is that the and worked ramp out new down, deals to of is been assembling want Typically, details specifications get X our next we human to several course next weeks. the few yet our are is signed requires the finalized, not ensure finalizing includes to and on to these with to several iterations. to yielding step the expecting Oftentimes, to the experience, systems, the Based put experts. believe place agreement engineers matter customer that the potential designs with work
will achieved This begins. X continue size X can X the refine including completed, gets factors, built X XX-month ramp-up throughout, our it end itself Ramp-up to full necessary advance. this test tough and typically. and we checkpoints knowable forecast. ramp-up on place. be These in take steady updating The to to the to being do are as of the dynamic will the roughly can But often of months state. which experience, model customer. can a taking to amount that ongoing that and so many achieve typically to a process upon with is of the slowly Once in up duration is ramp take the not months is with that depend based dependencies period. engagement tuning result quarterly typically This makes We and us customer
dramatically our quarter announcing about happened are revenue X There today, results weeks, from we company, there in last In was talk well worth a QX EBITDA revenue was in $X.X guidance. million, no after the let's QX last was management the from $XX.X media but $XXX,XXX. million the large which couple all second of contributed it of pulled also just adjusted in is in social Finally, that in the revenue quarter QX of no and back It knowing QX, deals and change. which was underwent half spending the in year, revenue year as significant last the close. of
that that this this QX in XXXX revenue out possible XXXX second does in our back account this our half, have for social we would not it is company, revenue but XX%. over business growth upside. media resume If from could the We believe quarter business plan been customer from large the
we social XXXX the high we EBITDA XXs, the we rate our expect revenue from exiting account annualized not does teens million also that rate Even discussed, be an company today. we run or the back potentially plan revenue could in have year from elements media if could for and of without basis. potentially X about growth our these out business on incorporated, XXXX $XX that exceed spoken just Our this new adjusted customers large
we cash numbers, the to a debt the no to short-term open questions. over line healthy with the million quarter will now then on and appreciable the sheet. ended and and Marissa the go We balance $XX.X I'll call in sheet, over balance investments for turn
